Let’s Play God

Let’s start today by answering a few nice chewy questions that some people have spent far too long worrying about.

  • Q: Did God create life? A: No.
  • Q: Is life a miracle? A: No.
  • Q: Is the creation of life on Earth a mystery? A: No.
  • Q: Was the creation of life an unlikely event? A: No.
  • Q: Can I have a go, preferably on my tea-break? A: Yes.

We can be utterly confident of these answers. Why? Because life is trivially easy to make and I’m going to show you how to do it. Then, if you want to play God and initiate Genesis on your laptop, all you’ll need to do is cut some code and hit run. As many times as you like, with as many variations as you like. You can spend your afternoon having Yahweh-happy-fun-times and see how far you get at reproducing Eden.

During my brief time at Princeton, there were a few simulation results that I built which really got me excited and this was one of them. It demonstrates that far from being mysterious or difficult to model, life can show up anywhere. All you need is a system that supports a suitably dense set of copying operations.

Before I explain what that means, I should first explain what I mean by creating life. It’s a pretty charged phrase with a lot of connotations. And what I’m not going to do is show you how to build Frankenstein’s monster in your living room. So for those who want to quibble over the implications of this result, this is where we get quibbly.

What I mean by life is a self-organizing, self-reproducing system that’s capable of adaptation. And what I mean by create is that this life will assemble itself from raw ingredients in its environment. And that those raw ingredients are in isolation not alive. I’m not talking about biochemistry here, and I’m not going to demonstrate any metabolic processes. This Genesis event will be entirely digital in nature, and very, very simple.

Some of you may wonder why, in that case, I imagine there’s anything new or special in this post. People have been messing around with artificial life since about 1993. The Tierra system has been used as the basis for numerous papers on artificial abiogenesis.

The answer is that I’ve never seen an artificial life system that boils down the requirements for life quite so much, or organizes so fast. The simulation I’m going to show you allows you to watch evolution, of a sort, in real time. What it proves is that, under the right conditions, life is an unavoidable consequence of thermodynamics. You can’t stop it from showing up. It’s more a matter of falling down stairs than inspired creation.

So how do you do create life-lite? Here’s how:

  1. Build a big grid of cells.
  2. Fill each cell with a randomly generated copying instruction. Each instruction should take the form: copy the instruction at position X relative to me to the cell at position Y. For instance, a cell might say, “Copy from 3 north and 2 left to 1 south and 4 right”.
  3. Pick a cell at random. Execute the instruction there, so long as it does not result in an instruction copying a copy of itself. If you pick a cell at the edge, copy from the opposite side of the grid, just as you might to wrap a computer game screen.
  4. Repeat step 3 until a single species has devoured all others.

Start running it and you get something like this. (I’ve colored the cells based on the instruction they contain to make the whole thing easy to see.)

It’s that simple. You can watch it evolving. Voila Genesis. But why does it work? After all, we’ve explicitly forbidden any instruction from copying itself.

It works because once patterns of instructions appear that mutually self-copy, they spread. And that means they take over from instructions that don’t spread. Another way of saying this is that in a system that’s dense in copying operations, self-replicating patterns are the entropic outcome. Furthermore, the simplest, fastest, most robust self-replicating patterns are likely to be the ones that dominate.

Clearly this simulation approach has limitations. The only kind of adaptation that can happen is when copying patterns start interfering with each other. There’s no true mutation. Furthermore, the descriptive limits on the instructions mean that life can only ever get so far. This simulation, sadly, never tries to take over the world.

But I find this digital petri-dish wonderful to watch in any case. It’s exciting to see the little digital critters duke it out for dominance. The result is different every time you watch. And, as a science fiction writer, I find this model powerfully suggestive. If it’s this easy to kick off self-replicating systems under the right conditions, have we underestimated the range of possible conditions where alien life might arise? And given that this kind of life is so easy to make, why isn’t it filling up our computers already? What is it about our existing digital environment that’s not adequately copy-rich?

To my mind, this simulation makes it clear that the real mystery on Earth isn’t the creation of life itself. Life starts the moment you create the right conditions. The real question is how those conditions arise in the first place. Understand that, and we might be able to reproduce them, both digitally and physically. Once we can do that, playing God won’t just be easy or fun to watch, it’ll be world-changing.

(My first novel, Roboteer, comes out from Gollancz on July 16th.)

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s